The CEO of Cadence said the company is undervalued. The stock barely moved. The market sees a software company with a 30% margin. It should see a tax on every AI chip produced. Check the inputs, ignore the hype.
Cadence is not a software company. It is an infrastructure tax. EDA tools are the picks and shovels of the semiconductor gold rush. Every chip—GPU, ASIC, TPU—requires Cadence's design flow. The market prices it as a cyclical software vendor. The logic is broken.
Context
Cadence holds ~30% of the global EDA market. Revenue in FY2024 was ~$46 billion. The broader EDA market is $150-180 billion, growing at 8-12% CAGR. But the leverage is staggering: every dollar of EDA revenue enables $200-300 of semiconductor output. The AI boom accelerates this. Cloud capex from the top four CSPs is projected to exceed $300 billion annually by 2027. A significant portion flows into custom chip design. Each design cycle consumes 20-30% of total cost in EDA tools and IP.
The CEO's argument is simple: the market underestimates Cadence's AI exposure. He is correct. But the reasons are deeper than he articulated.
Core: The Systematic Teardown
First, the leverage effect is mispriced. The market treats EDA as a cost center. In reality, it is a value multiplier. When a chip design fails, the EDA bill is a fraction of the loss. When it succeeds, the EDA vendor captures a tiny slice of the profit. This asymmetry means Cadence's revenue is a call option on chip design success. The market prices it as a linear function of design starts. It should be a convex function tied to total semiconductor value. The code was solid; the logic was not.
Second, the business model is shifting. Cadence is moving from license-based to platform-based revenue. The Cadence.AI platform embeds machine learning into the design flow. This is not a feature. It is a lock-in. Once a design team's workflows are optimized by AI-EDA, switching costs become prohibitive. The platform becomes the operating system for chip design. Market still classifies Cadence as application software. It should be classified as a platform with network effects. The error is systematic.
Third, the hidden information. The CEO's "undervalued" signal is a call for investors to re-evaluate the revenue durability. EDA revenue is sticky. Customer churn is near zero. The top 25 semiconductor companies are all Cadence clients. In my years auditing DeFi protocols, I learned that the infrastructure layer is often the last to be revalued. The same pattern holds for EDA. The market waits for a catalyst. The catalyst is already here: AI design starts are exploding. But the revenue lags by 18-24 months. A flat line in the stock price now is more dangerous than a spike. It means the market is ignoring the compounding demand.
Fourth, the risk of misclassification. Analysts compare Cadence to software peers like Adobe or Salesforce. The comparison is absurd. Cadence's customers are not enterprises buying licenses. They are chip designers who cannot operate without the tool. The dependency is absolute. The valuation multiple should be closer to a royalty stream than a subscription stream. The market is pricing it as if the tool can be replaced. It cannot—not in the short term. The switching cost is in the hundreds of millions and multiple years of engineering time.
Contrarian: What the Bulls Got Right
The bulls are correct about the AI tailwind. But they underestimate the risks. The biggest risk is not competition from Synopsys. It is the commoditization of EDA through open-source alternatives like Chisel and the rise of AI-driven design automation that could reduce the need for proprietary tools. Also, the China decoupling is a real drag. The Chinese market accounts for ~15% of revenue. If the U.S. expands export controls, that slice could shrink. Icebergs are not warnings; they are delays. The market sees the AI growth but ignores the structural headwinds from geopolitics and open-source encroachment. The bulls are right about the direction. They are wrong about the magnitude of the offset.
Takeaway
The market will eventually correct the classification error. The question is whether the correction comes before or after the next chip cycle. I am not betting on the timing. I am betting on the math. The math says Cadence is a tax on AI infrastructure. The market treats it as an expense. The difference is the opportunity.
